WIZ KID'S DIAL 911 BOOST: AI
Codes time-saving NYC dispatch tool
A Manhattan high schooler has designed an AI algorithm to help 911 callers get the aid they actually need, which would cut down on response times and eventually save cities millions, he told The Post.
Pierce Wright, 17, a soft-spoken junior at The Browning School on the Upper East Side, said his intricate model could assist emergency dispatchers by, for instance, predicting when a caller is enduring a mental-health episode.
“If the algorithm says, ‘I think it’s a mental-health call,’ then you can send a psychiatrist or a mental health professional with the EMS crew to assist the patient and provide the more appropriate care,” instead of simply rushing police to the scene, he said Wednesday.
“It’s saving time for the patient — and the city as well,” Wright said. “And it’s also able to free up an ambulance much faster.”
To design the algorithm, he combined his experience as an EMS worker with his data science prowess.
He spent the past year painstakingly coding the AI, then training it with nearly two decades’ worth of statistics gleaned from the city’s massive online database of about 24 million emergency calls.
His work paid off, he says, because his model can predict what resources are needed based solely on factors such as the incoming call’s ZIP code, time of day, police precinct and initial severity level.
The kicker? The model has an uncanny 94.5% success rate — 2.2% higher than its human counterparts, he said.
“You need very little input from the actual caller,” Wright said as he sat at a table in his family’s Upper East Side dining room, which overlooks Park Avenue.
“You can really just say, ‘This is what it looks like.’ And the 911 operator has the ZIP code, police precinct, time of day, etc.
“They put that all in there, and in a couple of seconds, the model outputs what it believes the call will be within about 94% accuracy.”
On-the-job experience
Wright said he was inspired by fielding calls on his own shifts as a volunteer EMS worker in Westport, Conn.
Many times, the crew was sent to respond to what turned out to be a mental health or substanceabuse call, not the true medical emergencies for which they are trained, he said. It’s also not helpful to the patient, who is often brought to a local emergency room and left to languish.
“That’s not providing care for the patient, and it’s also wasting the resources of the city,” he said, adding that his model would help eliminate patients “waiting at a hospital just to be let out.”
Mother Melanie marveled at her son’s ability to put in the mammoth amount of coding work.
“I was like, ‘I just I hope it works,’ ” she said with a laugh. “Because I would hate for him to feel like he put all this time into something, and it didn’t work!
“But it was so thrilling — seeing those light-bulb moments when he would have a breakthrough,” she added. “And that would carry him into the next phase.”
Wright said his program can also be used for other kinds of emergency calls, although he was quick to say it is meant to help dispatchers, not replace them.
But one day — after he makes it more customizable and accessible to the average person — the model could save cities millions of dollars and drastically cut down on response times, he said.
The project, which took about 200 hours to complete, has already earned him honors including first place at NYU’s April 7 TerraNYC STEM Fair.